On inf-convolution-based robust practical stabilization under computational uncertainty

This article is concerned with practical stabilization of nonlinear systems by means inf-convolution-based sample-and-hold control. It a fairly general technique based on generic nonsmooth control Lyapunov function (CLF) and robust to actuator uncertainty, measurement noise, etc. The itself involves computation descent directions the CLF. turns out that nonexact realization this leads not just quantitative, but also qualitative obstruction in sense result might fail be direction altogether there no straightforward way relate it direction. Disturbance, primarily complicate described issue even more. suggests modified w.r.t system as well computational uncertainty. assumptions CLF are mild, as, e.g., any piece-wise smooth function, which often results from numerical LF/CLF construction, satisfies them. A study three-wheel robot dynamical steering throttle under various tolerances w. r. t. uncertainty demonstrates relevance addressed necessity modifying used technique. Similar analyses may extended other methods involve optimization, such Dini aiming or steepest descent.